MSN search privacy and the government subpoena

Microsoft and Yahoo! have been much criticised in the wake of Google's refusal to hand over search data to the US government. They quietly handed over the data that was the subject of the subpoenas served on them over the summer last year. Microsoft's Ken Moss (General Manager – MSN Web Search) has explained a little more about their cooperation with the government at MSN Search's Weblog.

"Some facts have been reported, but mostly I’ve seen a ton of speculation reported as facts. I wanted to use this blog post to clarify some facts and to share with you what we are thinking here at MSN Search.

Let me start with this core principle statement: privacy of our customers is non-negotiable and something worth fighting to protect.

Now, on to the specifics.

Over the summer we were subpoenaed by the DOJ regarding a lawsuit. The subpoena requested that we produce data from our search service. We worked hard to scope the request to something that would be consistent with this principle. The applicable parties to the case received this data, and the parties agreed that the information specific to this case would remain confidential. Specifically, we produced a random sample of pages from our index and some aggregated query logs that listed queries and how often they occurred. Absolutely no personal data was involved.

With this data you:

CAN see how frequently some query terms occurred.
CANNOT look up an IP and see what they queried
CANNOT look for users who queried for both “TERM A” and “TERM B”.

At MSN Search, we have strict guidelines in place to protect the privacy of our customers data, and I think you’ll agree that privacy was fully protected. We tried to strike the right balance in a very sensitive matter.

Now that you have more information, you can be the judge."

If those commenting on the blog post are typical, the overwhelming judgement appears to be negative. Of course the readership of the blog will be fairly self selecting and not mecessarily representative but I'd have expected a clear dividing line to emerge between the anti-Microsoft and pro-Microsoft contingents. Unscientific I know but on the sample I skimmed 15 were anti and 3 pro.